Short Answer

An AI lab will solve another Millennium Prize Problem before 2028, driven by the announced AI-assisted proof of the Riemann Hypothesis that passed initial independent Lean code compilation and checks, an outcome priced at 90%.

1. Market Behavior & Drivers

The market's primary movement is tied to a claimed proof of the Riemann Hypothesis. On November 20, 2026, probabilities spiked 75 percentage points, from 12% to 87%, after Anthropic's Levent Alpöge announced he had found a proof with assistance from Claude AI. This single event accounts for the bulk of the market's upward repricing from its low starting base.
Other notable moves were also event-driven. A 12 percentage point spike on November 14 followed a Scientific American report that an OpenAI expert team had verified a separate, AI-generated proof for the Navier-Stokes equations. This reversed a decline from earlier in the month, including an 11 point drop on November 16, which is attributed to a "calendar-driven cascade unwind" from prior unconfirmed rumors. A subsequent 11 percentage point drop on November 21 lacks a clear driver in the provided materials. Sentiment in related markets reinforces the focus on the Riemann Hypothesis, which is priced at an 80% probability of being the next problem solved on Polymarket.
  • A resolution before 2028 remains highly probable following Anthropic's proof claim.
  • The period before July 1, 2027, provides ample time for independent verification.
  • Full verification before December 1, 2026, appears challenging given the short timeframe.

Who Wins and Why

Outcome Market Model Why
Before Dec 1, 2026 31.0% 26.9% Short timeframe before December 1, 2026, makes full independent verification highly challenging.
Before 2027 77.0% 77.3% Extended timeframe before January 1, 2027, offers greater opportunity for full independent verification.
Before Jul 1, 2027 88.0% 88.1% Significantly extended timeframe before July 1, 2027, allows ample opportunity for full independent verification.
Before 2028 90.0% 90.0% Substantially extended timeframe before January 1, 2028, allows for full independent verification and widespread recognition.

Current Context

Prediction markets reflect increased confidence in near-term AI Millennium Prize solutions. Kalshi's market "When will an AI lab solve another Millennium Prize Problem?" (listed September 12, 2026) prices "Before 2027" at 87-88% and "Before December 1, 2026" at 68%, up sharply by 73.5 percentage points in one day following the Riemann claim [1]. A related Polymarket market prices an AI lab announcing another solution by December 31, 2026, at 84% and by December 31, 2027, at 92% [2]. Polymarket's "Which Millennium Prize Problem will AI solve next?" market (ending January 1, 2028, with $484.5K volume) prices the Riemann Hypothesis as the next AI-solved problem at 80%, Hodge at 5%, BSD and Yang-Mills at 1% each, and P vs NP at less than 1% [3][2].
Recent AI advancements include multiple high-profile Millennium Problem proof claims. On November 20, 2026, Anthropic mathematician Levent Alpöge posted a four-page explanation and 17 million lines of Lean code, claiming a proof of the Riemann Hypothesis and Generalized Riemann Hypothesis developed with Claude over weeks; the AI-generated explanation reportedly totals over 250,000 pages [4][5][6][7]. MIT's Andrew Sutherland independently compiled the proof (175,187 Lean modules, zero errors, three standard axioms) and termed it 'an historic moment,' with nanoda kernel and con-leche checks passing by November 21, 2026 [7][8]. Simultaneously, Stanford postdoc Alessandro Masullo released a separate, fully human proof of the Generalized Riemann Hypothesis on November 20, 2026, endorsed by Stanford professor Ravi Vakil, creating parallel human and AI claims [5][1]. Previously, OpenAI announced on September 8, 2026, that thousands of AI agents produced a Navier-Stokes singularity proof in 88 hours, Lean-checked but controversial due to a 'forcing' component [9][10][11]. A month later, OpenAI released 722 AI-generated math manuscripts, and on October 28, it released a 1,147-page 4D Yang-Mills existence proof, validated in Lean by a committee led by Nobel laureate David Gross on November 10, 2026 [9][12]. As of November 21, 2026, the Clay Mathematics Institute still lists the Riemann Hypothesis as unsolved [8].
Prize award timelines remain distant despite proofs; expert forecasts vary widely. Clay Mathematics Institute rules require a qualifying refereed publication, at least two years elapsed, and general acceptance before a prize can be awarded, meaning a fully verified AI proof announced in late 2026 cannot win the $1 million prize before approximately 2028 [13][14][15][8]. Kalshi prices "Millennium Prize awarded for any unsolved problem before January 1, 2028" at 6-13%, and "Riemann wins the prize before 2028" at 6-9% [13][14]. Earlier LEAP forecasting research (August-September 2025) indicated median odds of 20% for AI solving another Millennium problem by end-2026 and 80% by 2030, with superforecasters at 85% by 2030 [16]. The Forecasting Research Institute noted these earlier forecasts (10% by end-2027) severely underestimated AI progress after the Navier-Stokes result [17]. OpenAI researcher Noam Brown had forecast a Millennium solve "not in 2026, probably not in 2027, maybe in 2028" on September 17, 2026, but later lost a bet against a 2027 solve; he expects approximately 3x acceleration in AI research [18]. Skeptics, like Daniel Litt, predict zero to one problems solved autonomously by AI in the next 10 years, noting the remaining five problems (Riemann, P vs NP, Hodge, BSD, Yang-Mills) are harder than Navier-Stokes [16][19]. In contrast, fast-progress forecasters suggest Hodge and BSD are "gimmes" and anticipate all remaining problems solved by AI before 2030 [16].
Sources (19)
  1. 1When will an AI lab solve another Millennium Prize Problem? Odds & Predictions 2027kalshi.com
  2. 2预测市场全景日报 | FLASH-FILLINGflash-filling.com
  3. 3Polymarketpolymarket.com
  4. 4Anthropic mathematician claims to have solved prestigious Millennium Prize problem with Claude | The Vergetheverge.com
  5. 5AI claimed a solution to the 167-year-old Riemann hypothesis — and a human did, too | TechCrunchtechcrunch.com
  6. 6AI may have solved math’s greatest mystery. Can humans understand the answer? | Science | AAASscience.org
  7. 7Riemann Hypothesis Proof Claim: Claimed vs Verified | explainx.ai Blogexplainx.ai
  8. 8Has the Riemann Hypothesis been solved? The Claude claimconvika.com
  9. 9Following the “mathocalypse,” mathematicians consider the future | MIT News | Massachusetts Institute of Technologynews.mit.edu
  10. 10The rise of AI in mathematics — and science | Nature Computational Sciencenature.com
  11. 11A millennium problem solved by AI? | plus.maths.orgplus.maths.org
  12. 12AI rivals tackle one of physics’ deepest mathematical challenges | Science | AAASscience.org
  13. 13Will a Millennium Prize be awarded for any unsolved Millennium Prize Problem before Jan 1, 2028? — Before 2028 | Forecasting Agentforecast.agenticlearning.ai
  14. 14Science & Tech prediction markets — prob.watchprob.watch
  15. 15Riemann hypothesis proof claim by an Anthropic mathematiciantrendandticker.com
  16. 16Probability AI Solves Another Millennium Prize Problem | LEAPleap.forecastingresearch.org
  17. 17How Accurate Have AI Progress Forecasts Been So Far? – Forecasting Research Instituteforecastingresearch.org
  18. 18Noam Brown expects faster AI research, but not an overnight explosionheardin.ai
  19. 19AI math capabilities could be jagged for a long time – Daniel Littyoutube.com

2. Price Chart

Historical Price (Probability)

Outcome probability
Date

3. Significant Price Movements

Notable price changes detected in the chart, along with research into what caused each movement.

Outcome: Before Dec 1, 2026

📉 November 21, 2026: 38.0pp drop

Price decreased from 69.0% to 31.0%

What happened: The provided research indicates a significant increase in the market price for the "Before Dec 1, 2026" outcome, rather than a 38.0 percentage point drop, with gains of +29pp to 31% and +66pp to 68% observed on November 21, 2026 [1]. The primary driver of this surge was the announcement on November 20, 2026, by Anthropic technical staff member Levent Alpöge, claiming a Claude-assisted proof of the Riemann Hypothesis and Generalized Riemann Hypothesis, a Millennium Prize Problem, formalized in Lean [2]. Alpöge also claimed to have resolved a loophole in OpenAI's Navier-Stokes solution [3]. This major scientific news, reported by traditional outlets, clearly led the market's repricing [2]. Social media activity, including a PTT.BEST post, rapidly disseminated these claims [4], acting as a contributing accelerant.

📉 November 08, 2026: 10.0pp drop

Price decreased from 12.0% to 2.0%

What happened: The provided research does not describe a 10.0 percentage point drop for the market "When will an AI lab solve another Millennium Prize Problem?" with the outcome "Before Dec 1, 2026" on November 08, 2026. Instead, the primary driver for a significant price increase in this market was Levent Alpöge's announcement on November 20, 2026, that he and Claude formally proved the Riemann Hypothesis with a large Lean code base [2][5]. This traditional news release led the "Before Dec 1, 2026" contract to surge from the low-30s to approximately 68% around November 20-21, 2026 [1][6]. Social media, specifically rapid independent verification of the Lean proof by multiple kernels and community members on platforms like Hacker News and Nitter [5][7], served as a contributing accelerant by quickly validating Alpöge's claim.

Outcome: Before 2027

📈 November 20, 2026: 75.0pp spike

Price increased from 12.0% to 87.0%

What happened: The 75.0 percentage point price spike on November 20, 2026, was primarily driven by Anthropic's Levent Alpöge's announcement that he had found a proof for the Riemann Hypothesis and Generalized Riemann Hypothesis with the assistance of Claude AI [8][2][9]. This significant claim directly addressed the market's focus on an AI lab solving a Millennium Prize Problem. Social media, especially X, acted as the immediate amplification channel, with figures like Harry Stoltz flagging the note at 13:49 UTC on November 20, quickly spreading awareness and appearing to coincide with the market surge [8][5]. Therefore, social media was a primary driver of this price movement.

📉 November 16, 2026: 11.0pp drop

Price decreased from 23.0% to 12.0%

What happened: The 11.0 percentage point drop on November 16, 2026, was primarily driven by a continuation of a calendar-driven cascade unwind, as the "Before 2027" contract steadily declined from its September peak [10][11]. Initial optimism, fueled by an X (Twitter) rumor from user 'leo' and an OpenAI statement of "substantial progress" in September, unwound without official verification or Clay Mathematics Institute recognition as the year-end deadline approached [12][11][13]. No specific social media activity or news event on or immediately prior to November 16, 2026, is identified as causing this particular drop; instead, the market was likely pricing in the diminishing time for resolution criteria to be met [11]. Social media was irrelevant to this specific drop, as later announcements (e.g., Alpöge's Riemann claim on November 20) spurred price increases [2][14][15][6].

📈 November 14, 2026: 12.0pp spike

Price increased from 11.0% to 23.0%

What happened: On November 14, 2026, the primary driver of the market price movement was the traditional news announcement that Scientific American reported OpenAI's appointed expert team had verified its September AI-generated Navier-Stokes proof [16]. This report directly coincided with the 12.0 percentage point spike in the "Before 2027" outcome on that day [15]. Although the Navier-Stokes solution was criticized as a loophole and formal acceptance requires years of scrutiny, the news of its verification by an expert team appears to have prompted the market's reaction [16]. Significant social media activity, such as Levent Alpöge's claim of an AI-aided Riemann Hypothesis proof, occurred later on November 20, 2026, making it irrelevant to this specific price move [2]. Therefore, social media was irrelevant for this particular movement.
Sources (16)
  1. 1Will any AI lab… before Dec 1, 2026? — 31.0% on Kalshipredictionbubbles.net
  2. 2Anthropic worker claims to have solved a Holy Grail of math problems with the help of AIapnews.com
  3. 3AI just solved the ‘real’ Navier-Stokes problem | Scientific Americanscientificamerican.com
  4. 4[爆卦] 見證歷史!Claude證明黎曼猜想+廣義黎曼 - 看板 Gossiping - PTT.BEST 批踢踢爆文ptt.best
  5. 5Levent Alpöge claims proof of the Generalized Riemann Hypothesis [pdf] | Hacker Newsnews.ycombinator.com
  6. 6Prediction Market Movers: Biggest Odds Changes Today | PredictionBubblespredictionbubbles.net
  7. 7SharewareFrom1994nitter.cf
  8. 8Riemann Hypothesis Solved? The Claude Proof Claim Explainedfelloai.com
  9. 9Anthropic may have solved major maths problem Reimann hypothesisbbc.com
  10. 10Millennium Prize AI Odds 2026: A 50-Point Unwind | Odsageodsage.com
  11. 11Will any AI lab solve a Millennium Prize Problem before 2027? | Forecasting Agentforecast.agenticlearning.ai
  12. 12OpenAI Millennium Prize Odds: 84¢ On An Unconfirmed Rumor | OddsShopperoddsshopper.com
  13. 13When will an AI lab solve another Millennium Prize Problem? - Kalshi Odds | CoinRithmcoinrithm.com
  14. 14Anthropic mathematician claims to have solved prestigious Millennium Prize problem with Claude | The Vergetheverge.com
  15. 15When will an AI lab solve another Millennium Prize Problem? Odds & Predictions 2027kalshi.com
  16. 16Ai Lab Millennium Prize Breakthrough Before Year Endroundtable.pubs.com

4. Market Data

Contract Snapshot

This Kalshi prediction market asks "When will an AI lab solve another Millennium Prize Problem?"

1. YES resolution: The market resolves "Yes" if an AI lab officially solves one of the Millennium Prize Problems and receives formal prize recognition by the specified contract deadline. 2. NO resolution: The market resolves "No" if an AI lab does not receive official prize recognition for solving a Millennium Prize Problem by the specified contract deadline. 3. Key dates/deadlines: Specific contracts are available with deadlines such as "Before Dec 1, 2026," "Before 2027" (implying Jan 1, 2027), and "Before Jul 1, 2027." The maximum payout date for at least one of these contracts is December 31, 2027. 4. Special settlement conditions: Official prize recognition is a requirement for settlement, which is noted to take "at least two years under official rules." The solution must originate from an "AI lab."

Available Contracts

Market options and current pricing

Outcome bucket Yes (price) No (price) Last trade probability
Before Dec 1, 2026 $0.57 $0.49 31%
Before 2027 $0.77 $0.24 77%
Before Jul 1, 2027 $0.91 $0.17 88%
Before 2028 $0.97 $0.12 90%

Market Discussion

Traders overwhelmingly anticipate an AI lab solving another Millennium Prize Problem by mid-2027 (88%), with strong sentiment (77%) for a solution before 2027, representing a significant increase in confidence over the past day. This is largely driven by news of an Anthropic mathematician, Levent Alpöge, claiming to have proven the Riemann Hypothesis with AI assistance, a claim supported by initial checks and independent mathematicians. Although official prize verification would take at least two years and a human has also independently posted a Riemann proof, the market heavily favors an AI-assisted breakthrough soon.

5. Trust Index

Octagon Trust Index Kalshi 56 Caution

Integrity risk· Resolution quality

How it adds up
Integrity80% of score63Caution
Trade quality20% of score31High Risk

Includes the cost to trade: a $10,000 order can't be filled here because the order book is too thin.

Trust score56Caution

Weighted blend with hard caps — a critically weak safety pillar, or a severe trading anomaly, caps the total regardless of the rest. Full methodology · About the Trust Index

Trust profile
Integrity4 screens run · 6 don't apply

6. What are the key verification milestones for Anthropic's Riemann Hypothesis proof, and which mathematical bodies' endorsements could significantly shift market odds before mid-2027?

Primary Verification MilestonesIndependent mathematician review without errors, top peer-reviewed journal publication, or official statement from recognized mathematical bodies [1]
Near-Term Verification TargetBefore mid-2027 [1]
Qualifying Endorsement BodiesAnnals of Mathematics, American Mathematical Society, International Mathematical Union, national academies of sciences, Nobel Foundation [2][3]
Verification milestones for the Riemann Hypothesis proof impact market odds. For Anthropic mathematician Levent Alpöge's Riemann Hypothesis proof, key verification milestones include successful independent mathematician review without identified errors, acceptance and publication in a top peer-reviewed journal, or an official statement from recognized mathematical bodies [1]. Any of these, or reports from two major Source Agencies confirming the resolution, would satisfy the Kalshi contract [1]. Such an event would likely push the "Before 2028" market odds for resolution near certainty [1].
Several near-term milestones could validate the proof before mid-2027. Before mid-2027, critical verification steps to monitor include the completion of independent mathematician review of Alpöge's proof with no identified errors [1]. Additionally, acceptance and publication in a top peer-reviewed journal, such as Annals of Mathematics, would be significant [1]. A written statement from influential bodies like the Clay Mathematics Institute (CMI), the American Mathematical Society (AMS), or the International Mathematical Union (IMU) officially recognizing the proof would also constitute a key milestone [1]. Reports from two major Source Agencies, such as AP or Reuters, specifically referencing this work would also count towards verification [1]. While formal CMI Millennium Prize recognition for any new claim is not possible before 2027, other market-relevant routes, such as peer-reviewed publication or a recognized-body statement, could materialize before mid-2027 [4].
Sources (4)
  1. 1When will an AI lab solve another Millennium Prize Problem?worldbyflow.com
  2. 2When will the Riemann Hypothesis be resolved? - Kalshi Odds | CoinRithmcoinrithm.com
  3. 3When will P vs NP be resolved? - Kalshi KXPNP-32 Odds | CoinRithmcoinrithm.com
  4. 4Will any AI lab solve a Millennium Prize Problem before 2027? | Forecasting Agentforecast.agenticlearning.ai

7. What are the primary technical and procedural hurdles that could prevent the recent AI proofs from Anthropic or OpenAI from being officially recognized before 2028?

Earliest CMI Publication (2026 claims)Late 2028 [1]
OpenAI Manuscripts Withdrawn3 of 719 due to sign error [2][3]
Manuscripts Lacking Lean Formalization419 of 719 (58%) [2][3]
Technical challenges present significant obstacles to recognizing AI-generated mathematical proofs. Key issues include mismatches in formalizations, such as between OpenAI's Navier-Stokes proof in Lean and Clay's official problem statement, where Palomar's registry explicitly notes it is not a certification of match [2]. Furthermore, while Lean verification was confirmed for the Yang-Mills existence proof, it did not establish the mass gap or the uniqueness of the infinite-volume limit, leaving the core problem unresolved [1]. Reliability is another considerable concern, evidenced by OpenAI withdrawing 3 of 719 manuscripts due to an error; additionally, 419 of 719 (58%) lacked Lean formalization, with any retraction or unresolved error ultimately voiding a market claim [2][3].
Procedural complexities also lengthen the timeline for official proof recognition. The extensive nature of AI proofs necessitates many months for real refereeing, which would make late 2028 the earliest possible publication date via the Clay Mathematics Institute (CMI) for a 2026 claim, thus preventing a Before-2028 resolution [1]. Moreover, Clay's prize rules were designed for human authors, requiring publication in a qualifying outlet and subsequent two years of "general acceptance in the global mathematics community" [2][4]. Additional hurdles include attribution disputes, which have arisen publicly concerning credit for Euler-equations solutions [2][3]. Finally, community and infrastructure bottlenecks, such as submission limits and a stated misalignment between AI companies' goals and the mathematical community's objectives, as expressed by 25 Fields medallists, further contribute to these challenges [2][5].
Sources (5)
  1. 1Will any AI lab solve a Millennium Prize Problem before 2027? | Forecasting Agentforecast.agenticlearning.ai
  2. 2The Referee Will See You Eventually - Scenaricascenarica.substack.com
  3. 3When will an AI lab solve another Millennium… Odds on Kalshi | DOXAdoxa.pro
  4. 4When will an AI lab solve another Millennium Prize Problem?worldbyflow.com
  5. 5When will an AI lab solve another Millennium Prize Problem? Odds & Predictions 2027kalshi.com

8. How do the AI methodologies used by OpenAI for Navier-Stokes/Yang-Mills and Anthropic for the Riemann Hypothesis compare in terms of architecture and potential for formal verification?

OpenAI Navier-Stokes Model CountApproximately 10,000 copies [1]
OpenAI Lean Formalizations (Oct 2026)300 (42%) of 719 manuscripts [1][2]
Anthropic Riemann Hypothesis LeadMathematician Levent Alpöge [3]
OpenAI used a distributed architecture to prove pure Yang-Mills existence. Their approach to complex mathematical problems, including Navier-Stokes and Yang-Mills, utilized a distributed architecture comprising approximately 10,000 models working in concert [1]. A secondary model was employed to translate the output into Lean, a formal proof assistant [1]. In their October 6, 2026 batch release, 300 (42%) of the 719 remaining manuscripts featured Lean formalizations [1][2]. A committee confirmed on November 9–10, 2026, that OpenAI’s Lean-checked construction successfully proved the existence of pure Yang–Mills [4].
Verification of AI-generated proofs presents challenges in matching meaning and logic. Despite the success, questions emerged regarding whether the Lean proof accurately matched the natural-language proof [4][1]. Formal verification analysis indicated that while Lean checking addresses the referee bottleneck for logic, it does not similarly resolve issues concerning meaning [4][1]. The formalization registry itself explicitly states that inclusion does not certify that a verified statement aligns precisely with its informal counterpart [1].
Anthropic's Claude assisted a mathematician in exploring the Riemann Hypothesis. In contrast, Anthropic's methodology for the Riemann Hypothesis involved its AI, Claude, serving as an assistant to mathematician Levent Alpöge [3]. Alpöge credited Claude with providing weeks of significant support for this endeavor [3]. Notably, Alpöge had previously collaborated with Tristan Buckmaster, leveraging AI and Lean to demonstrate the Euler equations blow up [1]. Information available on Anthropic's model does not detail specific architectural aspects beyond its function as an assistant [3]. As of mid-November 2026, mathematicians were actively evaluating Alpöge's four-page explanation of his proof [3].
Sources (4)
  1. 1The Referee Will See You Eventually - Scenaricascenarica.substack.com
  2. 2How many more Millennium Prize Problems will AI solve in 2026? | PredictPediapredictpedia.com
  3. 3When will an AI lab solve another Millennium Prize Problem? Odds & Predictions 2027kalshi.com
  4. 4Will any AI lab solve a Millennium Prize Problem before 2027? | Forecasting Agentforecast.agenticlearning.ai

9. What public code repositories and formal verification logs are available for tracking the compilation and checking of Anthropic's claimed Riemann Hypothesis proof?

Alpöge's proof explanation lengthFour-page explanation [1]
Claude's aid duration for Alpöge's proofWeeks [1]
Anthropic Fermat's Last Theorem formalization time30 hours [2]
Levent Alpöge's Riemann Hypothesis proof lacks public verification tracking. Mathematician Levent Alpöge's proof, reportedly aided by Claude for several weeks, does not have a dedicated public code repository or formal verification log available for tracking its compilation and checking [3]. Mathematicians are currently reviewing Alpöge's four-page explanation of his proof [1].
General Lean-based repositories exist, though none specifically track Alpöge's work. While there is no specific public tracking for Alpöge's proof, several Lean-based repositories exist for mathematical formalizations [3]. These include OpenAI's openai/math repository, which features a Lean formalization of the quasi-Riemann zero-free region Re(s)>7/8, the NavierStokesAndEuler Lean repository, Terence Tao's Palomar Lean-proof registry (opened August 18, 2026), and the Hexagon repository (announced October 6, 2026) [3]. It is important to note that Tao's registry explicitly states that listing does not certify that the formal statement matches the informal one [3]. The Hexagon repository accepts LLM-generated submissions at a rate of one paper per day [3].
Anthropic formalized Fermat's Last Theorem, but its Riemann result is partial. Anthropic also reported around August 2026 that approximately 1,000 Claude agents formalized Fermat's Last Theorem in Lean within 30 hours [2]. However, this was a formalization of an already known theorem, not a new mathematical result [2]. Anthropic's August 2026 Riemann result is currently considered a partial result and does not meet the criteria for a Millennium Prize Problem solution under Kalshi market rules [2].
Sources (3)
  1. 1When will an AI lab solve another Millennium Prize Problem? Odds & Predictions 2027kalshi.com
  2. 2Will any AI lab solve a Millennium Prize Problem before 2027? | Forecasting Agentforecast.agenticlearning.ai
  3. 3The Referee Will See You Eventually - Scenaricascenarica.substack.com

10. How do the verification pathways for Alessandro Masullo's human proof and Anthropic's AI-assisted proof of the Riemann Hypothesis compare in terms of timeline and community acceptance?

AI-assisted proof review statusStill reviewing as of late November 2026 [1]
Earliest Millennium Prize recognition2027-2028 [2][3]
Polymarket expectation for AI-solved Millennium ProblemRiemann Hypothesis 87% [4][5][6]
Two distinct Riemann Hypothesis proofs are currently undergoing review. One is a human-generated proof by Stanford postdoc Alessandro Masullo, and the other is an AI-assisted proof by Anthropic mathematician Levent Alpöge, which involved assistance from Claude over several weeks. As of late November 2026, the AI-assisted proof was still under review by mathematicians [1]. The specific review timeline for Masullo's human-generated proof is not detailed in the available information [1]. Under Clay Institute rules, formal prize recognition for any claimed solution from 2026 is not possible before 2027-2028, due to requirements including publication in a qualifying outlet, at least two years since publication, and "general acceptance in the global mathematics community" [2][3].
Prediction markets favor the AI-assisted proof for resolution. Despite the ongoing human verification processes, prediction markets suggest a higher likelihood that the AI-assisted proof will be the one to trigger a Millennium Prize Problem resolution, rather than Masullo's human proof [1][2][3][4][5][6]. For example, Polymarket shows 87% for the Riemann Hypothesis as the next Millennium Prize Problem to be solved by AI, and 74-76% for Anthropic announcing a Millennium Prize solution before January 1, 2027 [4][5][6].
AI-assisted proofs raise significant concerns within the mathematical community. The emergence of such proofs has sparked broader discussions among mathematicians. On September 11, 2026, 25 Fields medalists issued a statement expressing concerns that "the goals of the AI companies and the goals of the mathematical community are severely misaligned," with Terence Tao concluding an announcement with a plea: "do not abandon mathematics as a human activity" [2]. As of November 11, 2026, the available information shows only partial AI results for the Riemann, Hodge, and BSD problems, and no Millennium Problem has yet been officially resolved by an AI lab under market rules [3][1].
Sources (6)
  1. 1When will an AI lab solve another Millennium Prize Problem? Odds & Predictions 2027kalshi.com
  2. 2The Referee Will See You Eventually - Scenaricascenarica.substack.com
  3. 3Will any AI lab solve a Millennium Prize Problem before 2027? | Forecasting Agentforecast.agenticlearning.ai
  4. 4预测市场全景日报 | FLASH-FILLINGflash-filling.com
  5. 5Tech & AI Prediction Markets — Live Odds | Frenzy Capitalfrenzycap.com
  6. 6Which Millennium Prize Problem will AI solve next? · Riemann Hypothesis 87% — Polymarket oddspdata.world

11. What Could Change the Odds

Key Catalysts

Anthropic mathematician Levent Alpöge claimed proof of the Riemann hypothesis and Generalized Riemann Hypothesis (GRH) with Claude on November 20, 2026, releasing 17 million lines of Lean source files across 14 archives [1][2]. This claim followed Anthropic's Nasdaq IPO on November 19, 2026 [1][3]. Independent checkers, including MIT's Andrew Sutherland, confirmed the Lean code compiles, though full verification and a human-readable paper are pending [4]. Simultaneously, Stanford postdoc Alessandro Masullo posted a separate, fully human proof of the GRH [2]. Earlier AI milestones include OpenAI's September 8, 2026 Navier-Stokes proof, verified by its appointed review team by mid-November 2026, with the CMI stating on September 11 the problem "has apparently been settled" [5][6]. OpenAI also released 722+ math manuscripts in October [7], and GPT-5.2 Pro solved an Erdős problem in January 2026 [7].
Market pricing reflects a critical divergence in resolution criteria. Kalshi's 'Before 2028' market trades at approximately 66% [8], but its rules require a complete proof published in a reputable peer-reviewed journal, official CMI recognition, or two Source Agency reports; preprints, announcements, and machine-checked formalizations alone do not count [9]. Polymarket's announcement-based market, in contrast, prices 'AI lab announces another Millennium Prize solution by December 31, 2027' at 96% Yes [10]. This difference arises because Polymarket resolves on announcement, while Kalshi demands more stringent validation criteria [11]. The CMI's two-year post-publication scrutiny rule means the earliest any Clay prize consideration could occur is late 2028 [12]. Verification challenges persist, with a King's College/Cambridge preprint finding OpenAI's Lean code for Navier-Stokes diverged from its written proof [13], and Polymarket pricing a 2026 retraction at approximately 7% [11].

Key Dates & Catalysts

  • Expiration: October 07, 2026
  • Closes: January 01, 2028
Sources (13)
  1. 1Anthropic mathematician claims to have solved prestigious Millennium Prize problem with Claude | The Vergetheverge.com
  2. 2AI claimed a solution to the 167-year-old Riemann hypothesis — and a human did, too | TechCrunchtechcrunch.com
  3. 3Anthropic's Levent Alpöge says Claude helped prove Riemann hypothesisqz.com
  4. 4AI may have solved math’s greatest mystery. Can humans understand the answer? | Science | AAASscience.org
  5. 5Ai Lab Millennium Prize Breakthrough Before Year Endroundtable.pubs.com
  6. 6OpenAI Navier-Stokes Proof and the Next GPT Model | Intelsiintelsi.si
  7. 7What’s next if the Riemann hypothesis is solved? The road to verifying Alpöge and Claude’s proof claimclarin.com
  8. 8When will an AI lab solve another Millennium Prize Problem? (Before 2028): Odds, Chart & News | 4Search Predictions4search.com
  9. 9When will an AI lab solve another Millennium Prize Problem? - Kalshi Odds | CoinRithmcoinrithm.com
  10. 10Polymarketpolymarket.com
  11. 11When will an AI lab solve another Millennium Prize Problem? | PredictPediapredictpedia.com
  12. 12Polymarketpolymarket.com
  13. 13Who checks the proof? Lean, Navier-Stokes and the trust gapthedeepfeed.ai

13. Historical Resolutions

Historical Resolutions: 2 markets in this series

Outcomes: 0 resolved YES, 2 resolved NO

Recent resolutions:

  • KXAIMILLENNIUM-28-26OCT01: NO (Oct 01, 2026)
  • KXAIMILLENNIUM-28-26NOV01: NO (Nov 01, 2026)